Implementation and analysis of Flann (Fast Library for approximate Nearest neighbors) Implementation of OpenCV k-d tree

The k-d tree searches for the nearest point, using the Flann algorithm in OpenCV, which contains:

1: Achievements 2. Enquiry

See the procedure below:

#include "kdtree.h"
#include <iostream>
#include <iomanip>
#include "Cv.h"
#include "highgui.h"
#include <fstream>
#include "Cv.h"
#include "highgui.h"
#include <vector>//<vector > STL header File
#include <stdio.h>
#include "opencv2/objdetect/objdetect.hpp"
#include "opencv2/features2d/features2d.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/calib3d/calib3d.hpp"
#include "opencv2/imgproc/imgproc_c.h"
#include <iostream>


#define EXAMPLAR_NUM 700
#define Examplar_dim 3
The/**** system can search for the nearest point in 660 3-point clouds ****/
/************ Saving Cvmat matrix ****************************
Input parameters: &vecstring:n Point Cloud container
Output parameters: *matri: Convert to corresponding 3*n matrix
*************************************************************/
void Savecvmattotxt (const cvmat*matri, const char* filename)
{
Std::ofstream file (filename); Save File


for (int i = 0; i<matri->rows; i++)
{
for (int j = 0; j<matri->cols; j + +)
{
File << "" << cvmget (Matri, I, J);
}
File << Std::endl;
}
File.close ();
}
/*********** Read txt document save Cvmat matrix ****************************
Input parameters: &vecstring:n Point Cloud container
Output parameters: *matri: Convert to corresponding 3*n matrix
***************************************************************/
void Readtxttocvmat (Cvmat*matri, const char* filename)
{
Std::ifstream file (filename); Save File
Double temp;
for (int i = 0; i<matri->rows; i++)
{
for (int j = 0; j<matri->cols; j + +)
{
File >> temp;
Cvmset (Matri, I, j,temp);
}
}
File.close ();
}

int main (int argc, char *argv[])
{

Cv::mat Target (Ten, 3, cv_32f); Target Point Cloud (build k-d tree with it)

Readtxttocvmat (& (Cvmat) target, "data.txt");
std::cout << "target" <<target<< Std::endl;

1) Create Query tree: (Build k-d tree here)


Cv::flann::kdtreeindexparams Indexparams (4);//(this parameter is used to set the data structure of the build, where the k-d tree is selected)
Cv::flann::index Kdtree (target, indexparams);//Build k-d tree here with target


2) Find:
/*CV::P oint3f pt;
Std::cout << "Input Target point:" << Std::endl;
Std::cin >> pt.x >> pt.y >> pt.z;
std::vector<float> query;
Query.push_back (Pt.x);
Query.push_back (PT.Y);
Query.push_back (PT.Z); *//can be used to manually enter a single original point to be searched

Cv::mat source= Target; Original point
Cv::mat neibours (Source.rows, Source.cols, cv_32f);//Store The search to the point

int k = 1; Number of nearest neighbors

Cv::mat Indices (source.rows, K, cv_32f); The index of the corresponding point at which the search was loaded (that is, the number of rows of neibours in the target matrix)
Cv::mat dists (source.rows,k,cv_32f); Nearest nearest neighbor to search
Kdtree.knnsearch (source, indices, dists, K, Cv::flann::searchparams (32));


Std::cout << indices.at<int> (9, 0) << Std::endl;
Std::cout << dists.at<float> (9,0) << Std::endl;
for (int i = 0; i < neibours.rows; i++)
{
Neibours.row (i) = Target.Row (indices.at<int> (i, 0)) + 0;
}

Std::cout << "source=" << source << Std::endl;
Std::cout << "neibours" << neibours << Std::endl;
Std::cout << "indices=" << indices << Std::endl;
Std::cout << "dists=" << dists << Std::endl;


System ("PAUSE");
return 0;
}

Experimental results:

Implementation and analysis of Flann (Fast Library for approximate Nearest neighbors) Implementation of OpenCV k-d tree